The field of the invention relates generally to analyzing hyperspectral imagery, and more specifically, to designing and training hyperspectral systems.
Hyperspectral systems may be used for a large variety of missions, including, but not limited to, search and rescue, forestry, natural gas exploration, geology, agricultural analysis, and archeology. Prior to deployment, a hyperspectral system is trained to recognize items of interest in contrast to background or environment details. In many cases, this training is performed by having the hyperspectral system analyze a large plurality of hyperspectral images to learn how to differentiate pixels associated with items of interest from pixels associated with the background of the image. For example, a hyperspectral system may be trained to be able to recognize a tent in contrast to the surrounding forest. Depending on the mission, the hyperspectral system requires different training to recognize important features in contrast to background details. Proper training of a hyperspectral system may be expensive both in setting up and in training time. Furthermore, without proper design, the hyperspectral system may require additional training to meet the requirements of the mission.